X-Message-Number: 8029
Date: 09 Apr 97 22:54:27 EDT
From: "Steven B. Harris" <>
Subject: Surprise and the Subconscious

Dear Cryonet:

   Mike Perry writes:
   >>One way to resolve the apparent inconsistency is to think
about what "agent" is involved. Suppose, as a thought experiment,
that the brain of a person (not necessarily human, but a consci-
ous being) is divided into regions A and B that correspond to
different entities or "agents." This person communicates in
English, and generally behaves and functions enough like a human
that its consciousness can be tested in much the same way--plus
we have advanced technology as needed. It is found that (1)
region A is always active when the person is conscious, (2) when
A is inactive the person is never conscious, and (3) A is the
smallest possible region of the brain with these properties. So 
A, to all appearances, is the "seat of consciousness." 
   By analysis, though, it is found that B is very much like A,
except that B *only* communicates with A. When A is inactive, B
may be active, but it has no "hookup" with the outside world and
its activity is not normally apparent. In effect B is a fully
conscious entity in its own right, but it is not the "person"
that is represented in A. A uses B as a "slave"--with no direct
awareness of any "feelings" B may have.
  I'm no expert on the human brain, but it seems possible it
could have regions like B. I've heard of certain creative people
who can "work" on problems when they are not consciously aware 
of what they are doing, and they find answers. Perhaps they have 
"B-regions" that are actually conscious, to some degree, but it
isn't "their" consciousness. Following this train of thought, we
might imagine that much or all of the body's CNS is capable of
consciousness to some extent, yet it isn't the "seat of 
consciousness" to the person who inhabits that body. <<

-------------------------------------------------------------


   Comment: As Freud and Jung have pointed out, dream analysis
indicates that something like the above must be so.  When a
person dreams, they move through a world which is an imaginary
movie set, populated with "other people" who are also wholly
imaginary.  But who, then, is doing the imagining?  Not the
"dreamer," it seems, for the dreamer is frequently surprised (to
some extent-- more on this later) by things that happen in
dreams.  Obviously during a dream, the mind is "split" into at
least two parts.  One part is "doing" the simulation of reality
(like your own personal holodeck computer), and the other part is
doing the experiencing of that reality.  Thus for Freud the mind
comes in at least two parts-- a conscious, and a subconscious
that does imaginary simulations (a "part B").  The part B gets
input from the environment, but it doesn't need to.  Its output
is (at the very least) nonverbal.

   It is not clear whether the subconscious is "conscious" in the
usual Turing sense of being another functioning person (again,
consider the analogy of the holodeck computer-- it doesn't HAVE
to be).   But there is indirect evidence that the subconscious is
"conscious," albeit not very verbal, in much the same way that
there is evidence that nonverbal animals are consciousness.
During waking, the output of the subconscious is somewhat
suppressed by environmental stimuli, but there is no reason to
imagine that it is inactive.  So what is the subconscious doing,
during waking?  Some speculation follows.

   First of all, much experience indicates that the brain
possesses complex information-processing routines which are not
directly accessible to the conscious.  Simple examples are the
systems that controls physiologic balance (e.g., breathing), and
the higher-level systems that run motor behavior.  Some of these
are complex.  Indeed, most people have had the experience of
having their own bodies "drive them to work," or begin to, when
they didn't intend to go there, but had instead started out from
home without paying sufficient attention.  Of course, it is not
the body that drives one to work automatically in such cases, but
an automatic subprogram or subroutine (mental "agent," if you
will) that incorporates a lot of high-level tasks (visual
recognition, fine motor control, decision making at traffic
lights, etc, etc).  In other words, something *quite* 
sophisticated (so sophisticated that it is only now just becoming
a realistic goal with today's computer systems).  

   There is also evidence for subconscious information-processing
of a more purely cerebral nature.  It is possible (for example)
to show that information presented to the brain through the ears
while attention has been diverted, and which is not accessible to
the consciousness, nevertheless affects decision-making.  One
can, for instance, present the brain with two simultaneous 
conversations, only one of which can be "listened to," by the
conscious mind, at a time-- and only one of which is available on
formal testing.  But the information content of both 
conversations can be shown to be available on indirect testing. 
One must conclude from these tests that either the subject is
lying when he states that he doesn't know fact A which was
presented in a conversation he wasn't paying attention to (but
which can be shown to be influencing a test choice when he is
forced to "guess"), or else one must conclude that he actually
doesn't "know" the fact, but that "somebody else" DOES (i.e., his
subconscious mind, which doesn't do the speaking).

   Physically, where is the center of such "para-conscious"
information-processing?  In some experiments people are presented
patterns of symbols associated with an event (a good or bad
weather situation, for instance), and then forced, on the basis
of more symbols, to guess about a "forecast."   In these experim-
ents the information is shown to be processed partly in the basal
ganglia.  Normal people soon develop "intuitive" feelings about
what patterns of symbols predict, long before they can articulate
the relationships, whereas people with basal ganglia damage 
(Parkinson's disease patients) cannot do this.  So some of the
brain's "para-conscious" inductive pattern recognition is done at
the same physical level that controls higher level motor tasks
like driving (perhaps not surprisingly).  If anything, evidence
points to our basal ganglia (including the thalami) being the
seat of our intuitive "subconscious" minds (when it comes to
pattern recognition), but NOT our articulate, logical, conscious
minds (Although this too is simplistic, as diseases of the basal
ganglia like Huntington's disease cause severe dementia).

    At the same time, a well-known series of experiments suggests
that the right brain processes information independently of the
left brain, and presents this information to the left brain
somewhat inarticulately, in the form of emotional responses, and
also (perhaps) in more direct stimuli to act.  It is probable,
then, that some of the "subconscious" is located in the 
right-brain (perhaps in the right cortex, in somewhat the same
way that speech seems localized in the left cortex in most
people).

    What does the subconscious (which we've here referred to as
one entity, but which may of course be composed of many Minksi-an
"agents") ordinarily do, during waking?  Some of this has already
been alluded to.  Apparently the subconscious processes informa-
tion at many levels from the surroundings, and abstracts physical
patterns, emotional signals, and physical threats quite 
independently of "conscious" (by which I mean articulatable)
thought.  Its conclusions are then presented to the consciousness
in various ways, ranging from presenting (or hitting!) the
conscious mind with emotions (fear, anxiety, happiness, lust,
etc, etc), to causing in it an odd intuitive  "preference" for
certain actions, which is hard for the consciousness to rationa-
lly explain (hence the origin of irrational ESP and "spiritual
knowledge" myths).  It may even be that the subconscious (in the
form of one or more agents) has some direct control over motor
actions, and more directly influences physical behavior.

   If this happens, then here also the process may be difficult
for the "speaking consciousness" to explain (to put it mildly).  
For example, when a "murderer" says "I didn't mean to kill him--
the gun just went off," he may be speaking the literal truth
about how the situation appeared to the "mental agent" that is
(or "who" is) doing the talking.  Thus, a person judicially put
to death while protesting his innocence of a murder can be viewed
(somewhat macabrely) as a collection of mental agents, only some
of which are "guilty," and in many cases perhaps not including
the mental agent who is vocalizing innocence.  Understanding this
possibility is not very comforting, to be sure-- but then
understanding rarely is.

    The monolithic view of consciousness which prevailed before
Freud, did so partly because this view is somewhat "explicable"
(if such "explanations" count as explanations) in terms of
religious models.  In these models, those complex ideas and
behavioral proddings which arose in the conscious mind 
independently of consciousness, were explained as the influence
of devils or the spirit of the divine.  It took an atheistic
worldview to do better.  

   Development of a multi-agent view of consciousness has also in
the past been badly hampered by the fact that the mind seems to
be equipped with a prevarication or self-justification module (or
Munchausen Circuit, if you will) which causes the "speaking
consciousness" to quite often fabricate reasons for behavior. 
Some of these can be downright silly.  Experimenters in the early
commissurotomy right-brain/left-brain experiments, for instance,
were quite often shocked by the fact that the left brain had a
propensity to "explain" things that the right-brain/left hand
did, even when it was absolutely clear to the experimenters that
the explanation was bogus (due to the fact that the left brain
did not have the knowledge which caused the left hand to do what
it was doing).   So why did the left brain feel the need to
"explain" actions that it had no control over?  We still don't
know, but the horror is that such behavior **is not limited to
split-brain models.**  More recently, investigators have been
amazed to find that in normal intact persons the conscious mind
quite often gives completely bogus explanations for behaviors
which are the result of unconsciously processed information, and
that this lying behavior doesn't stop until the right brain is
anesthetized by an injection of barbiturate into the right
carotid.  Only then do people become honest, saying that they
don't know why they chose a certain test answer.  

   If this is the way our minds are set up, it is a wonder that
human beings have made any cultural or scientific progress at
all.  What a terrible handicap to honest communication!  (It's
enough to make one cynical).  One can only speculate about the
social evolutionary pressures which must have driven development
of such a formal spoken "modesty protector" for unconscious
drives, and for unconscious information-processing in general. 

   Is there anything else we can say about how the subconscious
processes information?  How the subconscious works to process
social information is again suggested by the content of dreams. 
In dreams, an entire "stage-show" is set up by one part of the
mind (or by a submind, is you like), and in it, characters are
allowed to talk and interact, as in life.  Dreams allow access of
the speaking consciousness to this stage show, but the most
interesting question is what happens to all of this when the
sleeper wakes.  Does the stage show disappear-- its only purpose
having been to entertain "us" when "we" are asleep?  This seems
unlikely.  

   A better explanation is that the "stage show" of dreams
actually runs all the time, but that in waking it is entrained by
inputs from the real world (which largely displace it in 
consciousness), so that the characters and social models are
continuously updated and allowed to run as simulations in a less
free way than in dreams (when allowed to diverge, such 
simulations are known as "daydreams").  How do we, as social
beings, know how another person is likely to react, if we say X
as opposed to Y?  Possibly the same way we know that a soccer-b-
all is going to go HERE rather than THERE if we hit it THIS way
rather than THAT way.  In all these cases, simulations have been
set up and run, largely but not completely beneath the level of
consciousness, and the results made available to us without any
of the messy calculatory details.   When the novelist jokes that
he needed to write another novel in a series in order to "find
out" what happened to the characters/people he cared about, such
a statement should be seen as only half specious.  Most, if not
all, novelists have had the feeling that characters initially
created have sometimes taken on a life of their own, balkily
refusing to say or do what the author had initially intended. 
From whence does such a decidedly odd (but almost universal)
feeling come?  Answer: it is entirely explainable in terms of
unconscious social modeling, something into which every 
successful story-writer must tap. 

   And also every successful politician, and indeed every
socially successful person.  Recently, much has been written
about "E.Q." or emotional I.Q.-- that mental ability which allows
a person to successfully model another person's internal state
and behavior.  People with low E.Q.s can be quite intelligent in
many formal ways but they are severely handicapped in social
situations.  (Examples are the nearly autistic physicist P.A.M.
Dirac and chemist Irving Langmuir, model for the socially
retarded scientist in Vonnegut's _Cat's Cradle_.)   Similarly
handicapped are many schizophrenics, who often suffer the same
devastating failure to be able to internally model other people's
feelings and behaviors (as well as other mental handicaps).  The
question of why humans have developed social cloaks for their
subconscious activities may be difficult, but the question of why
humans possess these activities in the first place is not. 
People without them have terrible problems interacting with
others.

   Are the one or more "agents" which make up the unconscious
mind, themselves "conscious"?  To this, it must be answered that
it (or they) have all the elements of consciousness, save speech. 
The subconscious is capable of interpreting spoken language, and
modeling complex social interactions.  To use a recent metaphor
from cryonet, it certainly "cringes" appropriately, and in shaded
and appropriate response to far more complex stimuli than would a
dog or even a chimp.  Thus, the subconscious is certainly
conscious in the ordinary sense, albeit mute.

                          ---


   Before winding up, I'd like to put in a few words about
solipsism.  For nearly the entire history of philosophy there has
been an undecidable running debate between objective realists and
idealists.  The objective realists hold that there exists an
objective external world outside consciousness, which "explains"
the phenomena of which consciousness becomes aware.  The 
idealists, on the other hand, point out that all phenomena which
we know anything of are (by definition) mental events, and thus
an objective world is in principle completely unknowable of
itself, and therefore the existence of objective reality is a
hypothesis which doesn't add anything to the question of why
events happen.

   Solipsists are egoistic persons who have taken idealism to its
logical conclusion, and decided that reality is nothing more than
their own mental reality, with nothing more than this existing. 
Solipsism is impossible to disprove, and the only thing one can
say about it, is that for all of that, it's hard for the average
person to keep pure.  The philosopher Bertrand Russell was once
entertained by a letter from a solipsist, who wrote that, since
solipsism was impossible to disprove, why then didn't more people
believe in solipsism as a philosophy?  Russell wrote back in his
incomparably ironic style, asking the correspondent why he
couldn't make himself *believe* that more people didn't believe
in solipsism....

   As Russell's comments suggest, one difficulty with solipsism
is the emotion of surprise.  Surprise results (as it did for
Russell's correspondent) when one's observations or sensations
don't correspond with what one expected, on the basis of one's
model of "reality."  Surprise is thus a fundamental emotion, and
perhaps a more important one than it is usually given credit for
being.  For the objective realist, surprise (which always results
from inductive mal-prediction) is the fundamental weeder-out of
bad inductive theories about the objective universe (c.f.
Popper's theories of science).  For the idealist, surprise is the
ultimate challenge to explain.  How does one account for stimuli
one is not expecting?  

   The solipsist, when confronted with surprise, is forced into
simple denial: the solipsist must either deny the existence of
surprise (clever tack), or else must simply hold somewhat
contrarily that it implies no contradiction in terms for one to
surprise one's self.  For the non-solipsist, however, the essence
of consciousness is knowledge, and the element of surprise in
unexpected stimuli to a conscious being is prima fasciae evidence
of the existence of *something* about which there has been a
demonstrated lack of predictive knowledge-- and thus, something
"outside" (by definition) that consciousness.  If solipsists
write letters to philosophers asking for explanations, shouldn't
they already know what to expect by way of answers?  

   To Rene Descartes' famous maxim (the so-called Cogito): "I
think, therefore I am," the non-solipsist will add a corollary:
"Furthermore, I'm often surprised, therefore there is more to
existence than just me."  The solipsist will, of course, at this
point object that the corollary is unfairly tautological, since
identity has been implicitly defined (arbitrarily) in terms of
knowledge, and this used to "deduce" existence of something other
than a single identity, from the fact of that identity's lack of
predictive knowledge.  But this is an unfair criticism, inasmuch
as all statements which are "logically true" or deductively true
(things that Kant would say we know by "pure reason") are
necessarily tautologies.  Descartes' _Cogito_ is itself a
tautology (at best, an "enthymeme," with a hidden premise), if
one looks closely.

   If the reader is still with me at this point, and hasn't gone
to sleep, it is possible to analyze some of the previous 
philosophical debate on Cryonet in terms of what the emotion of
surprise means to the average person.  At the beginning of this
letter, I made the explicit argument that the existence of
surprise in a dream argued for the existence of more than one
"model-producing" conscious entity in the brain of a dreaming
person.  At the same time, I can also as long time experimental
dreamer report that dreams are not as surprising as real-life
events.  In dreams, "objects" sometimes do surprising things, but
"people" only rarely do, and "people" never seen to act in
_totally_ inexplicable ways (as they do occasionally in real
life-- especially women <g>).  In dreams, I have noted that
characters act in ways that are understandable and somewhat
stereotyped, almost as people in a bad novel (perhaps this says
something about this author's creativity...).  Moreover, I have
noted that more than once, characters in my dreams seen to share
a shocking amount of my own knowledge about the world, so much so
that I've frequently awakened wishing that I might somewhere find
a real person who was as much an alter-ego as someone I had met
in a dream (no doubt if I actually did, I would be spectacularly
bored in a short time, of course).

   It's no coincidence that the characters in my dreams share my
knowledge, for we are dipping at the same well, so to speak. 
My dream characters thus share some of my identity, but not quite
all of it.  Empirically, one of the primary things that convinces
me of the separate reality/identity of other people, and of the
non-dream nature of the "waking world," is that in my waking
hours other people ("real people") do things which are utterly
unexpected.  I believe that (except for die-hard solipsists) we
accord identity to other people partly on this basis.  If other
people acted precisely as we expected them to, we would soon no
doubt come to see them as nothing more than complicated devices,
not deserving of the same status and regard with which we hold
ourselves.

   A machine which began to slavishly act in predictable ways in
a conversation would not pass the Turing test-- indeed, this is
the most damning of machine-like behaviors in such a test. 
Moreover, I propose that one reason Searle's "Chinese Room" seems
to be intrinsically incapable of consciousness to some imaginers,
is that the Chinese Room is intrinsically and by definition
incapable of surprising us.  You can, after all, look up and
predict anything it does, in a big book of rules.  For that
reason, it doesn't pass the Descartes Corollary test (the
"Surprise Test"), and neither do any computers which run with
discrete and predictable binary programs.  

   Here, I don't mean "predictable" in terms of the Turing
halting problem, I mean predicable from the view of repeatability
of behavior.  Humans would have exactly the same consciousness
problem, if they were found to be this kind of machine.  What
would happen if we were to make a matter duplicator which turned
out copies of people who always did exactly the same thing,
perhaps for hours or days on end, after they stepped off the
replicator pad, and into the test-room?  We might well begin to
regard humans in a different light, in that case.  Suppose I feel
"conscious," and yet am doing only exactly what I am expected and
predicted to do, just like an unconscious robot-- is that not a
paradox?  (Perhaps only a paradox of the sorites type?)

    Is such a duplicator and test-room even possible, then?  Some
people, after due consideration, would say no-- that different
conditions of the room at different times would preclude 
duplicate responses of the replicated human.  But what if you
replicate the room also?  Now many people will take refuge in
quantum mechanics, and argue that any such test-room will be
embedded in the universe, and will begin to diverge from any
duplicate (even a perfect one) which appears at another time and
place.  We've seen this argument on Cryonet.  Another way to
frame such an argument is to say that many people refuse to
accept the consciousness of any being found to be operating
according to fixed and completely deterministic rules, and thus
incapable of doing something we don't expect from looking at a
prior simulation.  If a being can't, in principle, provide a
surprise for one who knows enough about it, some people begin to
get solipsistic about its "consciousness."  Not to do so would be
to admit to the possibility of a certain determinism of action
for "conscious beings," which bothers many people.  As to whether
it should or not, I cannot say.*

                                  Steve Harris
   

* "Arguments" about determinism have some of the same odd 
character as arguments about solipsism.  If someone argues for
determinism, the rejoinder is: "What makes you say that..?"
If the argument is for some element of randomness, the rejoinder
is: "Oh, you're just saying that..."   People who want to argue a
third alternative need to be reminded, as Minski does, that there
isn't one.

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=8029